raster data and geodatabases for environmental management


Background: NASA, Wikimedia, 2012

introducing Raster Datasets

  • Data structure based on a matrix of numbers directly representing land cover
  • Can use a sparse matrix data format to save storage space
  • Tend to be used very often in environmental GIS applications
Background: NASA, Wikimedia, 2010

When read by a computer, the matrix values are mapped onto color schemes:







Remember, because it is a matrix, raster data can be processed by conducting operations on the cells of the matrix or by using  matrix algebra to conduct operations on the matrix as a whole.




Key examples of raster datasets:



national Land Cover Dataset

US Geological Survey, 2006:

http://www.mrlc.gov/nlcd2006.php



Lights at Night Datasets

NOAA, 2008:

http://sos.noaa.gov/Datasets/dataset.php?id=100


And, of course, the very interesting Distance to McDonald's dataset


Stephen Van Worley, Data Pointed, 2009:

http://www.datapointed.net/visualizations/maps/distance-to-nearest-mcdonalds/



Tons of interesting global raster datasets can be found at:

http://spatial-analyst.net/wiki/index.php?title=Global_datasets

Key raster terms

  • Pixel: the smallest part of a raster; an individual cell in the matrix


  • Resolution: the level of detail of a raster dataset, determined by the pixel size


  • Band: when the raster data are from remote sensing, the different wavelengths recorded

Modifiable Areal Unit Problem (MAUP)

  • Occurs when data are aggregated according to arbitrary areal units
      • In our case, pixels
  • Two forms:
    • Scale: different relationships visible depending on size of units
    • Zonation: different relationships appear depending on shape of units
  • Usually discussed for vector data; also a problem for raster because a pixel aggregates data



laboratory Exercise: Introducing Raster Manipulations with the National Land Cover Dataset

first, we would usually want to download the dataset from the site

(http://www.mrlc.gov/nlcd2006.php)


But the dataset is HUGE and we won't have enough time in the lab


Instead, we'll use a clipped version posted on the desire2learn site



Nevertheless, We'll go through the steps I took to get this cut down to a manageable chunk for the state of Illinois:

First, I grab our old friends the census tiger files to get an outline of illinois

(http://www.census.gov/geo/maps-data/data/tiger.html)

Then open up ArcMap and drop it in


Then use select >> select by attribute to get the state of Illinois

And export it as a separate shapefile

Making sure to save only the selected feature:

Then remove the US states layer and drop in the NLCD image (.img) file (be sure to unzip the file first!)




Remember to check the projections!  In this case, they're different, so we need to reproject the Illinois shapefile to match the NLCD image, which is in an Alber's equal area projection:

So open arc toolbox, choose data management tools >> projections and transformations >> feature >> project, and import the output coordinate system from the NLCD image before reprojecting:

Now we want to clip out a manageable bit of this behemoth of a file.  So we'll want to go to ArcToolbox >> Spatial Analyst Tools >> Extraction >> Extract by Mask, set our NLCD layer as our Input Raster, and use our reprojected Illinois shapefile as our mask


This is where we come in.  Add the IllinoisAlbers shapefile and the nlcdillinois image file to ArcMap.  What do you see?

So our first task will be to define the projection correctly.  This is done with ArcToolbox >> Projections and Transformations >> Raster >> Define Projection.  Set your Input Dataset as nlcdillinois and the coordinate system as Albers_Conical_Equal_Area

Well, that's better.  Now we just need to let ArcMap know that the values in this raster are codes (they represent land-use classes), not absolute values.  Open the Properties for the raster layer, go the Symbology tab, and set up a Discrete Color scheme.

Hmmm. . .  not too pretty, but we don't care that much.  Let's say that we're working on a project on the potential effects of fracking, and we're primarily concerned with forest cover, water, wetlands, and impervious surfaces.  It would be easier if we could just extract those values out of this raster.  




Let's look at the NLCD 2006 legend: http://www.mrlc.gov/nlcd06_leg.php.  What classes do we need?

We can split out these types using ArcToolbox >> Spatial Analyst Tools >> Reclass >> Reclassify, get unique values, and set our desired values to one and all other values to zero

Alternatively, we could use ArcToolbox >> Spatial Analyst Tools >> Map Algebra >> Raster Calculator.  You then create an expression like the following: nlcdillinois == 11.  This basically says "Give me a 1 where nlcdillinois and a 0 everywhere else."

On your own, repeat one of these processes to extract forests and wetlands.  Then you can remove your Illinois NLCD layer.  We won't need it anymore.




Add a Lab Journal entry

Because we're interested in fracking, we could also use some information about the geology of the state of Illinois.  Download the Illinois geology shapefile from the USGS website at: http://mrdata.usgs.gov/geology/state/, unzip it, and drop it in ArcMap

Check the Properties and perform any operations you need to in order to ensure the data are compatible

Now we want to figure out how much area in each of these geologies is water.  The easiest way to do this is to use ArcTools >> Spatial Analyst >> Zonal >> Tabulate Area.  Set your shapefile as your zones, with FID as your zone field and the water raster as your Input Value Raster.

 

That gives you a table with the water area for each feature.  You can join it to the original shapefile using the FID field.

Repeat this process to determine the forest area for each geological zone.




Add a Lab Journal entry








Thinking about Rasters

Let's take a moment to consider the benefits and limitations of raster data in environmental GIS applications

Land Use and Land Cover Change

 (LULCC)

  • Crucial research area in work on global environmental change
  • Major part of International Geosphere-Biosphere Program and the International Human Dimensions Program
  • Crucial for understanding urban footprint, suburbanization, etc.



Background: Hengl, IGBP, Wikimedia, 2004

What are the primary sources of LULCC data?


What is the tradeoff between coverage, cost, and land use data?


How does land cover extent affect the likelihood of having MAUPs?

What are temporal inconsistency problems?  Why are they an issue in lULCC data?


What are Spatial Inconsistency problems?


What is categorical uncertainty?

Why is the Land Use/Land Cover Distinction challenging?


How does integration help overcome some of the limitations of both raster and vector data?

Why is Validation important for LULCC data?


why is the quality of lULCC Data such an important issue for Climate Change modeling?

Geodatabases


  • Data structured used to store multiple layers in a single .gdb file
  • 3 types:
    • Personal: Essentially a Microsoft Access database; can be up to 2 GB but looses functionality between 250 and 500 MB; can only be edited by one person at a time
    • File: Multiple GIS datasets maintained in a single file folder; up to 1 TB in size; only one person can edit
    • ArcSDE: A relational database that uses an organization's database management system

typical Contents

  • Feature layers (like shapefiles)
  • Raster/image layers
  • Tables (usually attribute tables for feature layers)

Why Use it?

  • Single environment includes raster, vector, and tabular data
  • Can communicate with many other GIS formats
  • Can be scaled with the organization size and DBMS capacity
  • Can be used to establish relationships between datasets that make editing easier




Making a Geodatabase

Close ArcMap and open ArcCatalog.  Using the Catalog Tree, navigate to your working folder.  Right click on the folder, then select new >> file geodatabase

Now we need to add class into the geodatabase.  Right click on the geodatabase and choose import >> feature class >> single.  Choose one of your shapefiles and give it a reasonable name in the Output Feature Class field.

Now let's add a raster layer.  Right click on the geodatabase again and choose import >> Raster Datasets and select one of the raster datasets you created.

Now we'll import a table.  Right click on the geodatabase, choose import >> Table (single) and select one of the area tabulations you created



Import all the raster layers, Vector Layers, and Tables into your geodatabase, Including the Wells Layer




add a lab journal entry


Finally, Close ArcCatalog and Open ArcMap.  add in your geodatabase layers and make a map showing the location of wells, one of your landcover types, and the area of landcover of that type in each geological area in the state.  Take a screenshot of your data view.




Add a Lab Journal Entry

Made with Slides.com